Журнал

About Editorial Board Information for authors and reviewers Publication ethics Contacts User Account: send / review a manuscript

Improved delineation of contiguous agricultural fields in remote sensing image segmentation

R.V. Larionov¹ , V.V. Khryashchev¹ , A.L. Priorov¹

¹ P.G. Demidov Yaroslavl State University, Yaroslavl , Russia

Accepted: 17.04.2026
DOI: 10.21046/2070-7401-2026-23-4-51-62

The problem of automatic segmentation of agricultural lands in satellite images of remote sensing of the Earth is considered. The main focus of the research is aimed at solving the problem of separating adjacent objects with similar spectral characteristics, which is a common reason for merging several physically different fields into a single polygon using standard neural network approaches. Modern methods of field boundary allocation, including neural network architectures based on U-Net, are analyzed. To improve the quality of segmentation, a modified learning method is proposed, including adaptive balancing of training packages to eliminate data imbalance, the introduction of an additional class “narrow area between objects”, the use of a weight map for boundaries in the loss function, as well as the method of invariance to augmentation. Experiments on a set of images in the visible and near-infrared ranges from the Sentinel-2 satellite for the territory of the European part of Russia have shown that the combination of the proposed modifications makes it possible to increase the segmentation accuracy from 0.68 to 0.78 according to the Dice metric and from 0.39 to 0.51 according to the F1 detection metric, compared with the basic U-Net architecture. This confirms the ability of the proposed algorithm to correctly divide agricultural land into separate objects.

Keywords: remote sensing, image segmentation, farmland recognition, deep learning, convolutional neural network

Full text

References:

Amirova N. R., Sargina L. V., Kondratyeva Ya. E., Digital technologies in agriculture, CITISE, 2020, No. 2, pp. 266–280 (in Russian), DOI: 10.15350/2409-7616.2020.2.25.
Denisov P. V., Troshko K. A.., Lupyan E. A. et al., Satellite remote sensing for boundaries determination of agricultural lands, < i>Zemleustroistvo, kadastr i monitoring zemel’, 2024, V. 19, No. 4, pp. 227–232 (in Russian), DOI: 10.33920/sel-04-2404-04.
Larionov R. V., Sennikov A. V., Khryaschev V . V., Forest canopy changes modeling on the basis of remote surface sensing data, J. Automation in Industry, 2024, No. 2, pp. 32–35 (in Russian), DOI: 10.25728/avtprom.2024.02.09.
Nikolenko S., Kadurin A., Arkhangelskaya E ., Glubokoe obuchenie. Pogruzhenie v mir neironnykh setei (Deep learning. Dive into the world of neural networks), Saint Petersburg: Piter Publishing House, 2019, 480 p. (in Russian).
Skvortsov E. A., Skvortsova E. G., Sandu I. S., Iovlev G. A., Transition of agriculture to digital, intellectual and robotics technologies, Economy of Regions, 2018, No. 3, pp. 1014–1028 (in Russian), DOI: 10.17059/2018-3-23.
Tolpin V. A., Bartalev S. A., Elkina E. S. et al., The VEGA-GEOGLAM information system: a tool for the development of methods and approaches to using satellite remote sensing data in problem-solving tasks of global agricultural monitoring, Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2019, V. 16, No, 3, pp. 183–197 (in Russian), DOI: 10.21046/2070-7401-2019-16-3-183-197.
Troshko K. A., Denisov P. V., Dunaieva Ie. A. et al. (2025a), Development of winter crops in Russia in the spring-summer period of 2025 based on remote monitoring data, Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2025, V. 22, No. 3, pp. 321–328 (in Russian), DOI: 10.21046/2070-7401-2025-22-3-321-328.
Troshko K. A., Denisov P. V., Dunaieva Ie. A. et al. (2025b), Development of agricultural crops in Russia in 2025 based on remote monitoring data, Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2025, V. 22, No. 5, pp. 449–456 (in Russian), DOI: 10.21046/2070-7401-2025-22-5-449-456.
Fedorinov A. V., Volkov S. N., Denisov P. V. et al., Ustanovlenie granits zemel’ sel’skokhozyaistvennogo naznacheniya: opyt pilotnykh regionov (Establishing boundaries of agricultural lands: Experience of pilot regions), Moscow: RADUGA, 2023, 420 p. (in Russian).
Fedorinov A. V., Volkov S. N., Sorokina O. A. et al., Algoritm ustanovleniya granits kategorii zemel’ sel’skokhozyaistvennogo naznacheniya sel’skokhozyaistvennykh ugodii dlya tselei vovlecheniya zemel’ v khozyaistvennyi oborot (Algorithm for establishing the boundaries of agricultural land category for the purposes of involving land into economic circulation), Moscow: RADUGA, 2024, 224 p. (in Russian).
Belgiu M., Csillik O ., Sentinel-2 cropland mapping using pixel-based and object-based time-weighted dynamic time warping analysis, Remote Sensing of Environment, 2018, V. 204, pp. 509–523, DOI: 10.1016/j.rse.2017.10.005.
Gonsalez R. C., Woods R. E., Digital image processing, 4th ed., New York: Pearson Publisher, 2017, 1192 p.
Kamilaris A., Prenafeta-Boldú F. X., A review of the use of convolutional neural networks in agriculture, J. Agricultural Science, 2018, V. 156, No. 3, pp. 312–322, DOI: 10.1017/S0021859618000436.
Kamilaris A., Kartakoullis A., Prenafeta-Boldú F. X., A review on the practice of big data analysis in agriculture, Computers and Electronics in Agriculture, 2017, V. 143, pp. 23–37, DOI: 10.1016/j.compag.2017.09.037.
Kornbith S., Norouzi M., Lee H., Hinton G ., Similarity of neural network representations revisited, Proc. 36 th Intern. Conf. on Machine Learning (ICML), 2019, V. 97, pp. 3519–3529.
Ronneberger O., Fischer P., Brox T., U-Net: Convolutional networks for biomedical image segmentation, In: Medical Image Computing and Computer-Assisted Intervention — MICCAI 2015, Lecture Notes in Computer Science, V. 9351, Cham: Springer, 2015, pp. 234–241, DOI: 10.1007/978-3-319-24574-4_28.
Sedov A. G., Khryashchev V. V., Larionov R. V., Ostrovskaya A. A., Loss function selection in a problem of satellite image segmentation using convolutional neural network, 2019 Systems of Signal Synchronization, Generating and Processing in Telecommunications (SYNCHROINFO), 2019, 4 p., DOI: 10.1109/SYNCHROINFO.2019.8814279.
Tavera A., Arnaudo E., Masone C., Caputo B ., Augmentation invariance and adaptive sampling in semantic segmentation of agricultural aerial images, Proc. IEEE/CVF Conf. on Computer Vision and Pattern Recognition Workshops (CVPRW), 2022, pp. 1655-1664, DOI: 10.1109/CVPRW56347.2022.00172.
Waldner F., Diakogiannis F. I., Deep learning on edge: Extracting field boundaries from satellite images with a convolutional neural network, Remote Sensing of Environment, 2020, V. 245, Article 111741, DOI: 10.1016/j.rse.2020.111741.
Xu Y., Zhu Z., Guo M., Huang Y., Multiscale edge-guided network for accurate cultivated land parcel boundary extraction from remote sensing images, IEEE Trans. Geoscience and Remote Sensing, 2024, V. 62, Article 4501020, 20 p., DOI: 10.1109/TGRS.2023.3344122.

создание сайта - TXL